Is the content of the law necessarily grounded in moral considerations? While legal positivists answer “no” and anti-positivists answer “yes,” this disagreement until recently seemed to exist against a backdrop of agreement: both sides supposed that legal and moral norms were of distinct kinds.
A new wave of anti-positivist writing has challenged this assumption. These anti-positivists suggest that legal norms are not simply dependent on moral norms but are moral norms—more exactly, that they are that subset of moral norms that stand in some special relation to legal institutions. This striking thesis has rapidly come to represent the dominant view among anti-positivists. Following Ronald Dworkin, I shall refer to it as the one-system view of law.Footnote 1
This view no doubt inherits some of the strengths and weaknesses of other forms of anti-positivism, such as the interpretivism of Dworkin's Law's Empire.Footnote 2 But it is nevertheless a distinctive position that ought to be tested in its own right. One way to do this would be to assesses the theoretical considerations that are supposed to motivate it.Footnote 3 These considerations are varied. Dworkin's view in his late work was that the one-system view follows from the best interpretation of the concept of law, one “construct[ed] . . . by finding a justification of [legal] practices in a larger integrated network of political value.”Footnote 4 Scott Hershovitz, another proponent of the one-system view, is moved by a rejection of any sort of normativity or “quasi-normativity” in the legal domain that would admit the existence of norms that are not robustly normative in the way that morality is.Footnote 5 Mark Greenberg arrives at the one-system view partly due to skepticism of what he sees as the orthodox jurisprudential view that the content of the law is determined directly by officials’ authoritative pronouncements.Footnote 6 Other considerations, of course, motivate anti-positivism more generally: it is often claimed, for example, that anti-positivism can best account for the way that moral argument is deployed in legal cases.Footnote 7
As important as these background theoretical questions are, in this paper I focus on assessing the positive proposal these authors advance, and in particular on its extensional plausibility—on whether it renders tenable conclusions about legal content.
I will contend that it does not. I begin in Section I by setting out the structure of the one-system view and the strategy of my argument. Sections II and III develop the first of two criticisms: that the view cannot account for certain kinds of systematic divergence between legal and moral norms. I focus not just on obligations, but also on the wider range of normative incidents, such as privileges and powers,Footnote 8 to which the one-system view is supposed to extend.Footnote 9 (Where it is not important to distinguish among them, I sometimes use “norms” or just “obligations” to refer to these various incidents.) My second criticism, elaborated in Section IV, is that the one-system view's attempts to demarcate legal obligations within morality imply an under- or over-inclusive set of legal norms, or both. I respond to possible objections in Section V and reflect on the implications of my critique in Section VI.
I. INTRODUCTION
In his final major work, Justice for Hedgehogs, Ronald Dworkin embraced a new view of law: one sometimes glimpsed in his earlier writings,Footnote 10 but whose implications he had, by his own admission, not fully appreciated.Footnote 11 Whereas the orthodox, two-systems picture treats law and morality as different systems of norms, the new one-system view posits a single system of moral norms, within which legal norms are one subset. There are various kinds of moral rights, Dworkin says, some of which are political rights, and among political rights are those with a further special feature that makes them legal: “they are enforceable on demand in an adjudicative political institution such as a court.”Footnote 12
Whereas Dworkin's discussion is brief and suggestive, a detailed alternative has been developed by Mark Greenberg. According to his “Moral Impact Theory,” the content of the law—the set of legal rights, powers, etc.—is the set of moral rights, powers, etc., generated by the actions of legal institutions in “the legally proper way.”Footnote 13
Scott Hershovitz offers a different variant of the view. He too holds that legal obligations are “just a species of moral obligation.”Footnote 14 As we have seen, he also “denies that there is any distinctive domain of legal normativity to be determined.”Footnote 15 But that does not mean anything goes: we should, Hershovitz thinks, regard our legal obligations as a subset of our moral obligations. How we distinguish legal norms within morality—which subset we identify as legal—might depend on our practical purposes. Legal obligations might sometimes be best thought of as moral obligations whose source lies in certain institutional action, and sometimes as those enforceable in court.Footnote 16
The one-system view has other defenders. Jeremy Waldron has offered a sympathetic reconstruction and defense of Dworkin's account.Footnote 17 Nicos Stavropoulos has articulated a “pure” or “nonhybrid” interpretivism according to which legal rights and obligations are “moral rights and obligations that bear the right relation to institutional practice, which therefore government may enforce, and must do so on demand, through its institutions.”Footnote 18 And Stephen Schaus has proposed that legal rights are moral rights that legal institutions have the standing to hold us to.Footnote 19
It might be thought that the one-system view conflates the law as it is and as it ought to be. But its proponents accept and emphasize that the legal rights and obligations we have are frequently worse than those we might or should have had. That legal obligations can be morally suboptimal simply follows, on this view, from the fact that moral obligations in general can be morally suboptimal. That sounds paradoxical but is not. The idea is simply that morality is sensitive to an often-imperfect past, as where, regrettably, one makes two promises that cannot both be fulfilled; the obligations are not thereby canceled, and one must do one's best to get as close as possible to fulfilling both.Footnote 20 It may then be morally suboptimal that one has certain moral obligations—not because their existence is somehow a morally imperfect response to the facts as they are, but because it would be better, morally speaking, if those facts had not come to pass.
The one-system view's proponents suggest that this pattern pervades morality, including institutional morality, and therefore the morality of law. Poor judicial precedents and legislation can create expectations it would now be wrong to upset, or mean that some action would now amount to a failure to treat like cases alike. Democratic lawmaking might change the moral position even if the law enacted is regrettable. And statutes that solve coordination problems might generate obligations to go along with a scheme, even if another would have been preferable.
Despite the subtlety of these points, I will suggest that the one-system view remains vulnerable on several fronts. To see this, let us first clarify its core commitments. The convenience of the “one-system” label notwithstanding, it will not be important for our purposes whether morality and law are systemic in any interesting (or unified) sense. What matters is that any one-system view holds that legal norms (obligations, rights, etc.) are a subset of moral norms.Footnote 21 Its proponents therefore subscribe to:
Identity Thesis: Necessarily, every legal obligation (privilege, power, etc.) is a moral obligation (privilege, power, etc.).
The Identity Thesis asserts a metaphysical relationship between legal and moral norms: the former just are (a type of) the latter. It does not assert the reverse: not every moral obligation is a legal obligation. The modal operator “necessarily” ensures that the view does not simply make a claim about legal norms that happen now to exist.
The Identity Thesis implies a weaker extensional thesis:
Correspondence Thesis: Necessarily, a legal subject X has a legal obligation (privilege, power, etc.) to φ in circumstances C only if X has a moral obligation (privilege, power, etc.) to φ in C.
This says that wherever one has a legal obligation, there exists a moral obligation with the same content. If this were false, the Identity Thesis could not be true. For there could then be a legal obligation that corresponded to no moral obligation, and hence could not be a moral obligation.
Notice that asking whether legal and moral obligations correspond, in the relevant sense, does not assume they are separate entities whose content is independently knowable. For legal and moral obligations to correspond is just for them to have the same content. Every norm corresponds to itself, since it has the same content as itself. If legal obligations were moral obligations, they would correspond to them by being them. It is in this way open to the one-system view to make good on the Correspondence Thesis—as it must do to vindicate the Identity Thesis.
Any version of the one-system view needs two additional components. First, it must specify the force of the moral norms it counts as legal. One view, addressed in Section II, takes legal norms to be all-things-considered moral norms; another, discussed in Section III, takes them to be pro tanto moral norms. Second, any one-system view should include demarcation criteria to determine which moral obligations it counts as legal and which as nonlegal. These criteria, considered in Section IV, typically appeal either to the source of moral obligations or to their role as grounds of judicial enforcement.
In part of what follows, particularly Sections II and III, I attempt to show that the Identity Thesis is false by showing that the Correspondence Thesis, which it implies, is false.Footnote 22 In Section IV I argue that, however the Identity Thesis is paired with candidate demarcation criteria, the resulting view is unsustainable. The argument proceeds partly through considering examples. Here I need only suppose that there could be legal arrangements giving rise to the legal obligations I describe, though I hope that, even from the brief sketches I am able to provide, the reality and familiarity of such legal obligations in contemporary legal systems will be apparent. I assume the legal obligations I describe have come about as a result of ordinary types of law creation,Footnote 23 like legislation and judicial precedent, and that the legal position is clear in the circumstances, by whatever the relevant standards of interpretation in the jurisdiction are, so that if one denied that there was such a legal obligation, one would rightly be regarded by competent lawyers, judges, scholars, and so on as having erred.Footnote 24 And I will be concerned with the content of legal norms that obtain—with which “propositions of law”Footnote 25 are true—not simply with legal texts or utterances that give rise to those norms.Footnote 26 (My discussion will not extend to a thoroughly eliminativist positionFootnote 27 on which there is, or should be, no room for what Dworkin called the “doctrinal concept” of law;Footnote 28 that is, I will assume that there are facts of the matter about what legal rights and duties people have.Footnote 29)
My target is not anti-positivism in general but the one-system view in particular. That is important to emphasize, since opposing the one-system view is consistent with various jurisprudential positions. Someone who believes legal norms are necessarily grounded in moral considerations, for example, need not hold that legal norms just are, or have the same content as, moral norms.Footnote 30 Denying the Identity Thesis leaves open a range of views about how, if at all, morality figures in grounding legal content, whether there can be extremely unjust laws, whether legal reasoning requires moral reasoning, and so on.Footnote 31
Proponents of the one-system view sometimes accept one seemingly counterintuitive extensional consequence: that it is inconsistent with the existence of grossly unjust law, at least in one sense.Footnote 32 Yet the upshot of my argument is much wider: the one-system view misdescribes the legal position not only when the law is egregiously unjust or “truly evil,”Footnote 33 but also in many mundane and familiar contexts, both where the law is only somewhat morally defective, and where it is not morally defective at all. If the argument succeeds, it demonstrates a deep problem with the one-system view and constitutes, I believe, a weighty consideration against accepting it. Whether this consideration is thought decisive will depend on one's views about the methodology of jurisprudence—on how important extensional adequacy is for a theory of law—and that wider question is not one I can address here. I hope, in any event, to show the scale of the challenge facing the one-system view if it is to be made plausible.
II. THE ALL-THINGS-CONSIDERED VIEW
Obligations and other normative incidents can have an all-things-considered or merely pro tanto character.Footnote 34 We can approach this distinction by initially noting, in rough outline, some central features of ordinary moral obligations, which I take to be defeasible, i.e., pro tanto.Footnote 35 First, an obligation is something different from a simple reason to perform an action.Footnote 36 That is—and this leads to a second point—obligations do not just favor actions but at least presumptively require them. They tell us what we “have to” do or “must” do: they are mandatory.Footnote 37 Third, obligations have a special normative character when breached. The breach of an obligation is at least normally a wrong, and where the obligation is owed to someone in particular, the breach normally wrongs them.Footnote 38 Finally, obligations seem to place us in distinctive relations of interpersonal accountability: at least frequently, others may hold us to obligations, and their breach makes apt certain reactive attitudes (such as blame and guilt) and actions (such as apology and making amends).Footnote 39
An obligation understood along these or similar lines might nevertheless conflict with a contrary obligation (or perhaps reason) and be overridden; it is in this sense defeasible or pro tanto. A pro tanto obligation that is in fact defeated in particular circumstances we might think of as merely pro tanto. One that is not defeated is an all-things-considered obligation. In that respect an all-things-considered obligation specifies what one must do overall.Footnote 40 This may involve breaching a contrary pro tanto obligation: lying to prevent a friend's suffering distress, or breaking a promise to honor a more pressing commitment, might fulfill all-things-considered obligations but breach pro tanto ones. The defeated pro tanto obligations, though overridden, are not canceled in such cases: they leave some moral remainder, which explains why responses to their breach such as feeling guilty or apologizing will still tend to be apt. In order to assess the different versions of the one-system view, I will assume that—although it may be linguistically less natural—normative incidents other than obligations can also be understood as all-things-considered or pro tanto.Footnote 41 I will assume, for example, that there are all-things-considered powers and pro tanto powers, the latter defeasibly constituting the former, subject to there being no sufficiently strong pro tanto disabilities.
According to the all-things-considered version of the one-system view—ATC for short—legal obligations are certain all-things-considered moral obligations. By the Correspondence Thesis, it would follow that X has a legal obligation (privilege, power, etc.) to φ in C only if X has an all-things-considered moral obligation (privilege, power, etc.) to φ in C. Mark Greenberg subscribes explicitly to ATC,Footnote 42 and suggests that it explains why we treat the law “not merely as one relevant consideration among many, but as a central concern, indeed as excluding the relevance of other considerations.”Footnote 43
Yet we can easily imagine circumstances where one's all-things-considered moral and legal duties conflict. Take one of Greenberg's own examples, where you have an all-things-considered moral duty to take your suddenly ill mother to the hospital.Footnote 44 Those same circumstances might well give you an all-things-considered moral obligation to breach a legal obligation. If your mother is in terrible pain, which could be quickly alleviated at the hospital, and the roads are clear, you may have an all-things-considered obligation to exceed the speed limit on the way to the hospital, or take your brother's car, for which you are uninsured—and so on.Footnote 45 Since your all-things-considered moral obligation is to breach your legal obligations, the latter cannot be the former.Footnote 46
In other cases, legal obligations mandate what is usually or always morally wrongful. The law may require doctors to report undocumented immigrants for deportation, so that these patients are deterred from accessing important medical procedures; some doctors may have an all-things-considered moral duty not to log their patients’ immigration status, thereby encouraging treatment. Police officers may be required by their superiors to infiltrate and subvert harmless civil rights groups; the officers’ all-things-considered moral obligations might be to undermine these efforts by making their work ineffectual, tipping off the target to surveillance, or leaking to the press, thereby breaching obligations under their employment contracts and official secrecy laws. Many more cases with this structure can easily be generated. ATC is implausibly committed to denying the existence of such clear legal duties, since they do not correspond to all-things-considered moral duties.
Where one lacks a moral duty to contravene a legal duty, one might still have a moral privilege (aka a liberty) whether to do so. Over-criminalization—the criminal prohibition of conduct that is innocuous or else does not merit criminal sanction—is widespread. In some jurisdictions such overbreadth may be condemned by some constitutional doctrine, but not in all jurisdictions, and anyway not all the time. Consider statutes that criminalize mutually consensual sexual activity between minors, as is the position currently in English law.Footnote 47 As James Edwards observes, since the law “extends to kissing and touching through clothes . . . the formative sexual experiences of the vast majority of the population are now offences.”Footnote 48 Presumably such conduct is not in breach of all-things-considered moral duties. We might think of many more examples, from laws that criminalize harmless speech, to those prohibiting the viewing of some overbroad category of “terrorist material,” extending even to journalism or academic research, in which legal duties contrast with all-things-considered moral privileges. The moral stakes need not be high for this pattern to hold. Suppose that it is unlawful to enter or leave your local park except by a designated entrance or exit, but that on your trip home, as it is getting dark, you can take a shortcut through a gap in some railings and so avoid a long and unsafe walk through the park. It is difficult to believe that, all things considered, you never have a moral privilege to take the shortcut. If one has an all-things-considered moral privilege to not-φ, one lacks an all-things-considered moral duty to φ. ATC therefore fails to identify the legal duties (to φ) because there are no all-things-considered moral duties to which they could correspond.
In the converse situation, a legal privilege whether or not to φ contrasts with an all-things-considered moral duty to φ, as where legal regulation is overly lax. Consider legal privileges that allow companies to pollute excessively, banks to behave too riskily, manufacturers to not refund certain customers for defective goods, or employers to not pay a (sufficient) minimum wage. We can easily imagine that these persons are morally duty-bound to do what the law permits them not to do, so that they have no such all-things-considered moral privileges. By the Correspondence Thesis, the false conclusion that they have no such legal privileges either follows.
This result might be avoided with a different conception of legal privileges, on which a legal privilege to φ (e.g., pollute excessively) is simply the absence of a legal duty not to φ. There is a legal duty not to φ just if there is a moral duty to φ and this duty satisfies the one-system view's demarcation criteria. In the cases just considered, there is no such legal duty, since the moral duties not to φ (e.g., not to pollute excessively) are not enforceable in court, or properly generated by legal institutions. There is therefore no legal duty against φing, and this just is, on the present proposal, a legal privilege to φ.
This revised understanding of privileges may be more viable, but faces two difficulties.Footnote 49 First, it stands in need of justification, on pain of being ad hoc. In not requiring legal privileges to be moral privileges, it exempts privileges from the Identity and Correspondence Theses. But on what basis? Legal arrangements are complex webs of privileges, claim-rights, duties, powers, and so on. It is unclear why one component of these structures should rest on different explanatory foundations from the rest. Second, this response gives incorrect results in cases where the one-system view over-generates legal duties—as results, I argue in Section IV, from applying its demarcation criteria. Where the one-system view wrongly holds that there is some legal duty, the present understanding of privileges compounds the error, implying that there is no relevant legal privilege.
Consider finally divergence between all-things-considered moral powers and legal powers, as where legal powers are, morally speaking, excessively wide. A government minister might have a power, subject to no or very limited restrictions, to deny someone entry to the country, or deprive them of citizenship, where she deems doing so is “conducive to the public good.”Footnote 50 Employers might hold a legal power to fire employees at will (except on certain grounds such as race). It is plausible that such laws put morally unjustifiable power in the hands of ministers and bosses: morally, suppose, some greater measure of security for entrants, or citizens, or employees is required, so that the moral powers here are more circumscribed than the legal powers. Then the Correspondence Thesis does not hold, since these persons have a legal power to φ but a moral disability.
Notice that the response we considered in the case of privileges will not do here. The strategy there was to understand a legal incident not directly as the corresponding moral incident, but instead as the absence of the opposite legal incident. Applied here, it recommends asking not whether some person has a moral power, but whether they lack a legal disability. Asking this circumspect question would not give a different answer. An employer with the legal power to fire their employees at will also lacks the legal disability against doing so, since their decision will successfully change their employees’ legal position in the way it purports to.
Why does ATC give the wrong results across many kinds of case? One explanation is that all-things-considered rights and duties do not, so to speak, come cheap. What one is all-things-considered obliged to do often involves acting contrary to powerful countervailing reasons and pro tanto duties. Second and relatedly, ATC requires that a single group of moral obligations—those specially related to legal institutions—should turn out inevitably to have all-things-considered force. But that is not true of norms that comprise other similarly individuated departments of morality, such as promissory obligations or obligations of friendship.
Finally, all-things-considered moral incidents are highly context-dependent in a way that legal incidents are often not. A driver's all-things-considered moral obligations may depend on their driving competence, the moral urgency of a trip, and how clear the roads are, even when their legal obligation depends only on an invariant speed limit. Whether an employer has an all-things-considered moral power to fire their workers depends on a mix of moral facts (about, for example, domination and exploitation) and empirical facts (about, for example, labor market flexibility) that may be irrelevant to their legal powers. It is not that the law cannot or does not ever account for moral considerations or contextual variability; it frequently does so via mechanisms such as reasonableness standards. The point is that when the law is not sensitive to context in this way—as where it sets a numerical speed limit—one's all-things-considered moral obligations may remain highly context-sensitive. This is a deep contrast in the structure of legal and moral obligation.
III. THE PRO TANTO VIEW
According to PT, the pro tanto variant of the one-system view, legal obligations are certain pro tanto moral obligations. Though he eschews the “pro tanto” terminology, this is in substance Hershovitz's view, given how we are here understanding pro tanto obligations.Footnote 51 By the Correspondence Thesis, PT holds that X has a legal obligation (privilege, power, etc.) to φ in C only if X has a pro tanto moral obligation (privilege, power, etc.) to φ in C.
PT can avoid some of the problems afflicting ATC. Pro tanto incidents are less scarce than all-things-considered ones, and PT can better account for legal and moral conflict, treating some legal obligations as moral obligations that are defeated by countervailing all-things-considered moral obligations.
Still, PT must identify legal obligations with genuine—albeit defeasible—moral obligations. Recall from Section II that even pro tanto obligations are notably constraining. They operate differently from mere reasons by (presumptively) requiring action, so that breaching an obligation is apt to constitute a wrong.
Putting things this way foregrounds the stringency of obligations, but as I alluded to in Section II, on some views what is distinctive about obligations is how they embed duty-bearers in schemes of interpersonal accountability.Footnote 52 To have an obligation, on this picture, amounts to (or implies) others’ having standing to hold one to some standard, which, if breached, licenses reactive attitudes on their part and that of the wrongdoer, as well as, typically, other secondary consequences, such as apology or perhaps compensation.
This view does not ease the burden of establishing moral obligations, however. Accountability-focused theories do not deny that obligations are also structured differently from ordinary reasonsFootnote 53—requiring, not simply favoring, action—and that they are standardly wrongful to breach.Footnote 54 Indeed, accountability theories are frequently defended on the basis that they help to explain why obligations are stringent in these ways.Footnote 55 Moreover, the relevant relations of accountability such theories emphasize are a serious matter. Having an obligation, on such views, means others are entitled to demand things of us,Footnote 56 while wrongdoers properly feel guilty and inspire blame in others.Footnote 57 These practices and attitudes are not trivial moral phenomena.
Notice finally that, on any view, particular obligations cannot be inferred from generalizations about obligations. Promises might generally generate pro tanto obligations, but it does not follow that a promise to rob a bank does so. More prosaically: a pro tanto obligation of politeness might generally apply, but perhaps not to people who are first victims of unprovoked rudeness. It is not that pro tanto obligations arise in such cases but are outweighed; sometimes, in the particular circumstances, obligations just do not arise at all.
With these points in mind, let us consider whether PT can explain certain types of case that ATC could not. Start with legal duties that conflict with moral duties. One might try first to vindicate legal duties as pro tanto moral duties by appealing to “content-dependent” moral considerations, those that refer to the value of the act one is putatively obliged to perform. This strategy, however, is not very plausible. Think again of simple cases where one has some pressing need to contravene the rules of the road. Plausibly one very often has pro tanto obligations to comply with these rules. But it is difficult to imagine that, for every driver under all possible circumstances, this will be true—difficult to believe, for example, that, on an obviously deserted country road with no danger present, an experienced driver invariably violates an obligation in driving before the traffic lights have turned to green, or not indicating before a turn.Footnote 58 At any rate, it is not easy to see the case for such obligations by focusing on the proscribed action. By hypothesis, such cases involve no risk: there seems to be no wrong at issue, no one who could reasonably feel aggrieved, no apt sense of guilt. Think too of cases where a legal duty ought to be resisted, as where the police officer ought to frustrate rather than further his department's attempts to subvert civil rights groups. Focusing on the legally required action makes it obscure why there should be even a pro tanto duty here—why we should think the officer does something even in one way wrong, or justifying legitimate guilt or blame,Footnote 59 if he frustrates the subversion. No doubt single actors in complex systems are sometimes imperfect judges of their actions’ significance, but it is implausible to insist that they can never see that they would be wrongful. Problems also seem to persist where the law's duties contrast with moral privileges. Where the law criminalizes quite harmless conduct, for example, there will be no content-dependent reason to think the legal prohibition will correspond to a moral one. After all, there is, in a banal sense, nothing wrong with the conduct in question.
These examples suggest that PT cannot be defended by appeal only to content-dependent moral considerations, so it will have to also appeal to “content-independent” considerations, such as democracy, fairness, solving coordination problems, and the vindication of legitimate expectations.Footnote 60 Yet, because each of these considerations is importantly limited in scope, they will not apply in a sufficiently reliable way to vindicate the Correspondence Thesis. These considerations, first, are often inapplicable and, second, where they do apply, result in a mismatch between moral and legal duties.
Consider the moral duty not to upset others’ expectations. This will only apply where certain conditions are met—for example, where someone has an expectation that would be upset, where the expectation is reasonable, and where the putative duty-bearer should bear the burden of meeting the expectation (because, for instance, they were responsible for causing it). Coordination problems, meanwhile, are not always relevantly at issue. Where they are, failing to conform to a legal obligation does not always threaten to unsettle the coordinating scheme, or threaten to do so with sufficient probability to generate a pro tanto duty to conform (as with speeding on a deserted road, or taking a shortcut through an empty park). What is more, nonconformity is often neutral or preferable from the point of view of coordination. As Daniel Viehoff has argued, many possible coordination solutions are always close to the one legally enacted.Footnote 61 The law may institute one scheme, S, requiring you to φ, but your not-φing might bring about a move to a new scheme, S*, which is like S in all respects except for your (or some small number's) not-φing, and that is as valuable as S or better.Footnote 62 So from the point of view of coordination, you might have good reason to perform, and certainly no pro tanto duty not to perform, acts that are wrongfully prohibited (as in the cases of over-criminalization or where one should resist a wrongful legal duty), or just prohibited by over-general rules (like the rules of the road).Footnote 63
Fairness, understood as something like treating like cases alike or as principled consistency, may likewise be inapplicable, or fail to ground a duty. It will be relevant only where there are at least two persons or groups sufficiently similarly situated such that there is a prospect of unfair treatment as between them. Even this is not always the case, since some legal norms can apply to single entities (such as a government agency). More significantly, if a law has already been applied inconsistently or arbitrarily, or has been under-enforced, fairness may already be compromised, and whether a norm is applied to a new case may be neutral from the point of view of fairness.
Aside from cases of outright inapplicability, fairness is also subject to problems of mismatch between the content of legal and moral norms. Fairness might matter variously in the law: axiologically (fairer states of affairs are more valuable), or in terms of reasons or duties to bring about fairer states of affairs. Greenberg notes that “the fact that a case is resolved in a particular way provides a reason for treating relevantly similar cases in the same way in the future.”Footnote 64 But it is a further task to explain how not just reasons but duties are generated, and generated systematically enough to do much work on PT. A judge might decline to treat like cases alike by confining a previous similar—but entirely misguided—decision “to its facts” rather than applying it to a new case. Even if there would be something to be said in favor of being consistent—some value or reason served by it—it seems doubtful that, by limiting the damage of the first case, the judge does something even in one way wrong, or meriting of indignation, blame, or guilt.Footnote 65
Moreover, accounts of fairness or principled consistency in the law are generally addressed to judges and other officials as the putative duty-bearers: since the court must treat X and Y alike, and the court required X to φ, it must now require Y to φ.Footnote 66 But these claims about the permissibility or obligatoriness of official enforcement do not imply that Y, a subject of the law, has a duty of fairness to φ, especially where φing would be a wrong. Even supposing that everyone, Y included, has an obligation to avoid unfair states of affairs, it would follow only that Y should not resist the norm's enforcement against her. But this does not show that she has a moral duty of consistency or fairness to φ, for her not-φing would not alter the fact that everyone is in the same legal-institutional position: subject to the same requirement to φ on pain of sanction. Her φing is not itself required to sustain whatever fairness inheres in the consistent application of or subjection to legal norms.
Consider next the directionality of these duties. Assume that A has injured B, and the law imposes some morally objectionable kind of liability on A, say by holding him responsible for, and requiring him to compensate, certain unforeseeable harms. A fairness-based explanation of a corresponding moral duty on A would appeal to this legal directive's having been applied to others in A's position. But fairness-based explanations struggle to explain how A's legal duty is owed only to B, whom he has injured.
Suppose, first, that fairness gives the consistent application of norms a kind of derivative moral force. This would wrongly suggest that A's legal duty is to state actors, such as courts, who have applied the norm consistently. Alternatively, one may think unfairness is a matter of differences in how persons fare, so that its disvalue is impersonal. This suggests, however, that A's duty is nondirected, i.e., not owed to anyone in particular, and does not explain its B-directedness. A third possibility is that A owes the duty on pain of unfairness to those who have had (or will have) to bear the duty in similar circumstances. But if so, the duty would seem to be owed to those finding themselves in A-like circumstances, rather than to B. It is true, of course, that B also loses out if an exception is made for A, but the mere fact that B has an interest in A's φing does not put A under a duty to φ, still less a duty of fairness specifically. Even if B were among the comparative victims, as it were, of A's being freed of this legal obligation, A's duty of fairness would be owed widely: both to B and to others in A's position.Footnote 67 If A breaches his duty, this explanation would count A as having wronged all in this wide class. Yet that is not the legal position: A's legal duty is owed distinctively to B, and only B is capable of being wronged in law by A's failure to φ.
Finally, consider the moral relevance of democratic authorization. It is, of course, highly controversial whether democratic laws ever, ipso facto, generate duties of conformity. But even if we allow arguendo that they can in principle, much law is not covered by such democratic considerations. That is obviously so where lawmaking institutions are straightforwardly undemocratic—not a trivial matter, since a sizeable proportion of present nation-states are, and the overwhelming majority of legal systems in history were, not democracies. Many more are, and were, partial or flawed democracies. And even in what are often regarded as paradigms of contemporary democratic politics, realism is required. Here, too, we frequently find gerrymandered constituencies; grossly disproportionate electoral systems; representatives beholden to wealthy corporate donors; voter suppression; policies that are democratically unaccountable because they are shielded by official secrecy laws; and laws subject to no or trivial legislative scrutiny. It is a difficult question when precisely such circumstances undercut a polity's ability to generate pro tanto duties to conform to legal directives. But even proponents of the authority of democracy accept that its authority is limited under certain real-world conditions, including but not only conditions such as these.Footnote 68 Finally, even in fully functioning democracies, much law—such as that made by courts or executive agencies—is not democratically made, and so does not yield obligations of conformity on democratic grounds.
What is the upshot of these considerations? The basic problem is that PT seeks to explain pro tanto moral (and hence legal) duties via a patchwork of moral mechanisms. Since each mechanism is importantly limited in scope, there is no general reason to believe that some or other will, for any clear legal obligation, apply so as to ensure there is a corresponding pro tanto moral obligation. Indeed, the patchwork is bound to be incomplete. It will always be possible to find cases where no moral mechanism applies to generate a corresponding duty. For we can always choose cases in which a legal duty obtains and persists even as we subtract the circumstances that would be needed to generate the corresponding pro tanto moral duty. In the clearest possible case, we can think of legal norms created nondemocratically (say by appointed judges or a flawed legislature) or that impose duties on nonvoters (such as noncitizens or minors), enforced inconsistently, to do things that are, in the circumstances, pointless or wrong, and not required to sustain valuable coordination. Then arguments from democracy, fairness, reasonable expectations, coordination, and the like will not have meaningful purchase. Of course, this is to put the point at its most cautious. Even in cases where one or more of these considerations are engaged, problems of duty mismatch will persist, where moral obligations have a different content to the legal obligations. We have already seen how these problems affect a range of examples.
Greenberg holds that it is a virtue of the one-system view that it allows that different considerations, which individually would be insufficient to generate a moral duty, can together combine to do so.Footnote 69 This possibility no doubt exists, but does little to motivate the idea that reasons will systematically combine in this way. The mere addition of moral reasons does not transform one or more of them into a duty. And, if each consideration applies to some limited class of cases, those cases in which one or more considerations overlap, and do so with sufficient moral heft to generate a duty, will similarly be a limited class.
Now, one might try to ground PT not in a patchwork of moral considerations but by still more wide-ranging arguments—such as those from fair play, consent, or associative obligation—that purport to establish a fully general pro tanto obligation to obey legal directives. It would be futile to try to survey them all here; the difficulties such arguments face in establishing such a general duty have been widely argued for.Footnote 70 It is in that context that the promise of the one-system view appeared to lie in not needing such a general account, but relying instead on a combination of more localized moral considerations. Yet as we have now also seen, this strategy yields an incomplete account of our legal obligations.
We have been focusing on obligations for simplicity. But consider now privileges. Why would corporations legally permitted to pollute excessively, or to pay derisory wages to their employees, have pro tanto moral privileges to commit such wrongs? Their having pro tanto privileges is unnecessary from the perspective of ensuring the success of the scheme: polluting below the legal limit, or paying higher wages, would not threaten it. And any democratic decision—or such corporations’ legitimate expectations—would concern how far their conduct is regulated by state institutions, not its general moral status.
A proponent of PT might adopt the strategy we considered in the case of ATC of exempting privileges from the Identity Thesis, so that a legal privilege does not imply a moral privilege. We need not repeat that discussion. Instead, I want to note a different possible response, which treats the relevant legal privileges as moral privileges held against the state.Footnote 71 Perhaps for one to have a pro tanto moral privilege to φ, in a way that is also a legal privilege, is just for the state to have no right that one not-φ. Thus, if the state has not enacted a minimum wage statute, it may have no moral right against employers that they pay fairer wages; employers would therefore hold a privilege against the state to not pay more.
Treating this relationship with the state as sufficient to explain the employer's legal privilege distorts it directionality. In law, the employer's privilege not to pay higher wages is held against the employee. It is equivalent to the absence of a directed duty owed to the employee to pay them a higher wage, and means the employee lacks a directed legal right against the employer to be paid more. None of this is captured in the employer's relation with the state. The distorting effects of a state-directed view of legal incidents are clearer still if applied to duties. It implies that an employer who fails to pay an employee breaches a duty owed to—and legally wrongs—the state, not the employee.
Moving, then, from privileges to powers, PT has fewer options open to it. Even if PT can exempt privileges from the Identity and Correspondence Theses and construe them as the mere absence of legal duties, we saw in Section II that this move does not help for powers. And here, just as with obligations, there is no reason to think that where a legal power exists some moral mechanism will always ground a relevant pro tanto moral power. Return to the employer with excessively broad legal powers to fire employees. One might think that arguments from democracy and coordination morally require conformity to a whole scheme—say of employment law—including the powers it confers. Many of the doubts already canvassed, particularly concerning the limits of coordination, would then apply. Moreover, this suggestion would at most license the conclusion that others have an obligation to treat the legal power–holder as if they had a moral power. It would not follow that they in fact had the moral power in question.Footnote 72
In sum, although PT is on stronger ground than ATC, it is also unable to sustain the Correspondence Thesis. Proponents of the one-system view sometimes claim that, although their position will not, as Greenberg puts it, countenance “truly evil law,” it will “explain much morally flawed law.”Footnote 73 Waldron contrasts “law-creating events that morality frowns upon, or law-creating events whose moral significance is a matter of dispute among us” with “downright evil laws,” the implication seemingly being that those in the latter category pose a particular explanatory difficulty for the theory.Footnote 74 Hershovitz says that explaining legal norms that conflict with moral duties is “straightforward enough,” whereas “[t]he tricky cases are the ones in which it looks like we have a legal obligation to do something morally repugnant.”Footnote 75 We are now in a position to see why these assessments are over-optimistic. Hardly any—perhaps none—of the examples we considered, which one or both of ATC and PT could not explain, rose to the level of the truly heinous. Many were, to be sure, instances of morally bad law—but of a sort familiar even in notionally liberal democratic polities. And they were not idiosyncratic: as we saw, they have structural features that are shared by countless other legal norms.
Focusing on such cases demonstrates the problems with the one-system view even in legal systems most favorable to its thesis, without needing to consider examples from Apartheid South Africa or the Antebellum South. The view cannot therefore be defended by a restriction of scope of the kind recently proposed by Greenberg to “a theoretically interesting class of systems that includes . . . for example, the United Stated and the United Kingdom,”Footnote 76 since the argument holds even for this more limited class of legal systems. (It is unclear, and Greenberg does not say, what unites these systems as “theoretically interesting.” It would be surprising if the very nature of legal obligation should change as one crossed borders, or within a jurisdiction at different points in its history.)
Nor can the one-system view be defended by merely allowing that legal texts, utterances, or other actions of legal officials, can be morally defective or worthy of criticism.Footnote 77 The argument here concerns the view's verdicts about the legal norms grounded at least partly in such texts and institutional action. That the one-system view's conclusions are plausible on certain other fronts is no response to the charge that they are implausible in this central respect. It is not as though we have a certain quantity of moral criticism that we are determined to be able to express of law-like things, so that a theory that provides avenues for expressing it somehow or other is defensible.
The explanatory problem for the one-system view is not just, or even mainly, about grossly unjust law. It extends much more widely; we cannot assume that, just because a legal duty is not a moral outrage, there is bound to be some moral explanation or other at work sufficient to ground a corresponding moral duty. Indeed, it is important to see that the failure of the Correspondence Thesis is not fundamentally a matter of morally defective laws at all. Because legal norms are highly general, even morally optimal laws will tend to cover particular circumstances for particular persons where the balance of moral considerations favors or permits nonconformity. In these cases, there might nonetheless be a pro tanto duty to stick to what the law requires, but as we have seen, that will often not be the case.
IV. DEMARCATION CRITERIA
There is a second group of extensional problems for the one-system view. These arise from the demarcation criteria it employs to explain which moral obligations are legal and which are nonlegal.
One might object right away that this question of demarcation is unimportant. Distinguishing within morality between legal obligations and other moral obligations might seem comparable to wondering what precisely makes something a “family obligation” or “work obligation.”Footnote 78 This skepticism about our question would be misplaced, however. For one thing, it is not a general truth that distinctions between species of moral obligation do not matter. We might, for instance, evaluate a theory of promising in part by how it distinguishes promissory obligations from other, nonvoluntary obligations. Since divergence between legal and (other) moral obligations pervades our experience of law, it is an important part of the explanandum for a theory of law. More fundamentally, this skepticism about the importance of intra-morality distinctions applies only if legal obligations are indeed moral obligations. It therefore assumes that we have already accepted the one-system view. But that gets things backwards: the view's demarcation criteria—since they partly determine its verdicts about when legal obligations exist—bear directly on whether to accept it in the first place. As we will see, the verdicts these criteria yield give us further reason to doubt it.
The importance of the demarcation question is not diminished by Hershovitz's suggestion that different demarcations may be appropriate depending on our practical purposes. That is not only because we may doubt that legal claims really are context-sensitive in this way.Footnote 79 Even assuming that they are, it would be surprising if there were not, at a minimum, a range of standard contexts—judges deciding cases, citizens considering their rights, professors teaching students—across which some core of legal propositions held in common, reflecting common demarcation criteria.Footnote 80 And even if there were little or nothing common to different contexts for demarcating legal norms, we would still need to assess the plausibility of any given demarcation. It does not matter how many demarcations are possible if none supports tenable verdicts about legal content.
What, then, are the one-system view's strongest candidates for how legal norms should be demarcated? There are broadly two proposals. The first holds that moral norms are legal just if they are appropriately generated by the actions of legal institutions. It is most strongly associated with Mark Greenberg's Moral Impact Theory, and I will call it MI for short. It has also been endorsed by Scott Hershovitz as one possible way to demarcate legal norms.Footnote 81 The second view holds that moral norms are legal just if they are legally enforceable—that is, enforceable in courts or similar adjudicative institutions. This view has been defended by Dworkin,Footnote 82 Hershovitz,Footnote 83 and Schaus.Footnote 84 Call it LE.
MI holds that for a moral obligation to be legal, it must have been part of the moral impact of the actions of legal institutions, such as courts and legislatures. Greenberg understands “impact” capaciously: moral obligations still count as generated, in the relevant sense, if they are “altered or reinforced” by legal-institutional action.Footnote 85 More significantly, such moral obligations must have been generated in what Greenberg calls the “legally proper way.”Footnote 86
Greenberg acknowledges that he lacks a complete account of this last idea. Although we cannot rule out the possibility that it could be made to work, there are strong reasons to doubt it. To be viable it must be independently plausible and explanatory—“not ad hoc,” as Greenberg maintains.Footnote 87 Yet it is not easy to pin down, in part because he defines it negatively. A moral obligation is not generated in the legally proper way insofar as: (a) “legal institutional action, by making the moral situation worse, generates obligations to remedy, oppose, or otherwise mitigate the consequences of the action”;Footnote 88 or (b) “an institution . . . explicitly purports not to be generating binding obligations”Footnote 89 (as where a solution is explicitly suggested but not mandated by a legal institution); or (c) “legal institutions’ actions lead through an extended chain of events to moral obligations that are intuitively too far downstream to qualify as legal obligations”Footnote 90 (as where a promise to help someone fill out their tax return, and hence the attendant moral obligation, is caused, very remotely, by the passing of tax laws). These negative conditions are said to flow from our “intuitive understanding of the way in which legal systems are supposed to generate obligations” and the idea that “a legal system, by its nature, is supposed to change the moral situation for the better,”Footnote 91 including (vis-à-vis (b)) “by generating all-things-considered binding obligations.”Footnote 92
Greenberg's suggestion that there is intuitive support for the idea of “the legally proper way” to change moral obligations is doubtful. The notion is a technical one, that people do not think or talk about as such; nor is it clear that we have tacit views about it. We do, of course, confidently judge that in the circumstances described in (a)–(c), the putative legal obligations have not been created. When we hear of a promise to help a friend with their tax return, for example, we unsurprisingly judge that the promise did not create a legal obligation. But that does not imply that we operate with some notion of the legally proper way of changing obligations—still less moral obligations—or that we reach judgments about legal obligations in virtue of such an underlying idea. The natural verdicts about (a)–(c) are compatible with many views about how legal content can be generated. They are, for example, easily explained within a broadly positivist picture: the posited law—such things as legislation and judicial precedent—will not in most jurisdictions pick out the relevant moral obligations as legal. Laws, as ordinarily interpreted, will not include in their content obligations to resist them; explicitly nonbinding language will not tend to count as law-creating; and unanticipated obligations, only remotely causally related to those contemplated in a legal rule, will not generally be part of the content of that norm. Equally, an interpretivist can readily say that the best interpretation of the legal materials will not lend itself to such obligations having been generated. Greenberg's conditions (a)–(c) do not, therefore, provide positive support for any distinctive idea of the “legally proper way” to change obligations.
Does the idea that legal systems are “supposed to change the moral situation for the better” take things further in understanding the “legally proper way”? If so, it does not take us the whole way. Greenberg treats cases like (c), where the moral obligations are too causally remote from acts of lawmaking, as sufficient to make the creation of obligations not legally proper—even though it might well be morally better that these obligations (e.g., to help friends with their tax returns) have arisen. So condition (c) cannot be justified by reference to improving the moral situation.
Because the notion of the legally proper way of creating obligations is introduced without fully articulated criteria for its application, it is sometimes unclear on what basis it is satisfied. Greenberg imagines a case where a statute's “linguistic meaning,” on any view of linguistic meaning, “clearly designates a particular scheme,” Scheme A, but where Scheme B becomes salient, “perhaps because a private company that is a dominant player in the relevant industry misinterprets the statute early on, or because a widespread psychological tendency makes it unlikely that people would adhere to scheme A.”Footnote 93 Greenberg assumes that there are moral reasons going both ways, but stipulates that the “overall effect is to generate an obligation to participate in B, not A.”Footnote 94 Since he supposes that this obligation, which involves flouting the clear linguistic meaning of the statute, is a legal obligation, he must take it to have been generated in the legally proper way.
This example further undercuts the claim that we have an intuitive grasp on the appropriate way of creating moral-legal obligations. For it seems not at all intuitive that legal institutions that mean to bring about one legal obligation instead, via an unforeseen causal chain, properly bring about a contrary one. Greenberg may be supposing that, while there can be several explanations of what makes a change to obligations legally proper, “changing the moral situation for the better” is sufficient. This would explain the obligation to participate in B. But this is not plausible, since it would then follow that, whenever the law caused people to take on valuable commitments within legal frameworks, the attendant moral duties would thereby become legal duties, including in cases like (c), if the downstream obligations are morally valuable.
If making the moral situation better is not sufficient to make a moral impact legally proper, it might be necessary. Yet this interpretation would seem to make MI under-inclusive. “Better” here must presumably be construed not against the baseline of a society without law at all (as the condition for legal properness would then always be trivially satisfied) but against the law existing before the relevant change. But if so, whenever a change in the law is for the worse, since the law does not do what it is supposed to do, it does not create obligations in the legally proper way, and to that extent fails to generate legal obligations. The condition would then not be satisfied where institutions implement a scheme that is inferior to the previous one, and people have obligations to go along with it (as plausibly happens at least sometimes when, say, a morally worse tax code is introduced). Yet it cannot be true that, just on this basis, the institutions failed to make law. Greenberg agrees, since he suggests that California's Proposition 13 made the tax code morally worse but still generated legal obligations in the legally proper way.Footnote 95
A more general reason to suppose MI will misidentify legal duties concerns legal discretions, where someone has a legal power and a privilege whether to φ (where φing involves changing others’ legal positions). Consider the discretion some highest courts possess whether to overrule their own previous case law, an administrative agency's discretion to implement one of several policies, or the discretion held by a discretionary trustee over how to distribute property among beneficiaries.Footnote 96 In such cases there will often be a moral obligation, not mandated by the law itself, to take a particular course of action—to overrule, pursue some administrative policy, or favor some distribution of trust property. Not only are these moral obligations brought about by the actions of legal institutions, the choice situations in which they arise—judicial decisions, discretionary trusts, etc.—are themselves the creations of the law. If the creation of these moral obligations is for the better, it is unclear on what basis MI could resist concluding incorrectly that they are legal.
Let us then turn to LE, which holds that moral obligations are legal if they are enforceable in court. Significantly, while “enforceable” will be useful shorthand, the relevant notion here is moral, not descriptive. For Dworkin, legal rights are those that people are “entitled to enforce on demand, without further legislative intervention.”Footnote 97 Schaus, meanwhile, suggests that legal obligations are moral obligations marked by the “distinctive liabilities we incur for violating them”Footnote 98 and that courts have the “moral standing to hold us to.”Footnote 99
However exactly LE understands enforceability, in this moralized sense, its mistake is to tie legal norms too closely to institutional mechanisms for adjudicating legal disputes. Some problems of the court-centricity of LE have been noted by Lawrence Sager, who argues that certain duties of American constitutional law—concerning, for instance, social rights—may be breached without being properly adjudicable in court.Footnote 100 Dworkin denies that these duties are legal; thinking otherwise, he says, assumes the two-systems picture that he rejects.Footnote 101 I consider such objections in Section V. For now, let us see how deeply the problem of tying legal rights to adjudication mechanisms affects LE.
A natural way to understand LE is:
(1) A norm, N, is a legal norm in circumstances C if and only if it is a moral norm that is judicially enforceable in C.
However, this would falsely imply that a legal right is extinguished when a procedural or evidential rule properly prevents a court from adjudicating it. A may not be able to enforce a claim against B because A cannot file within the time limit, A cannot afford court fees, there is insufficient evidence, the claim is not yet “ripe,” or the claim has become “moot”—and so forth. These doctrines block enforcement but do not cancel the underlying right. After all, A's right is not that [B φ, conditional on A's bringing a procedurally valid claim with sufficient evidence]. A's right is just that B φ. If B fails to φ, B without more commits the legal wrong, and could not claim to have acted lawfully on the basis that A did not bring a valid or sufficiently evidenced claim.
A proponent of LE should therefore accept that for A's moral right to be legal, it need not be enforceable in fact; it suffices if it is, as Schaus puts it, enforceable “in principle.”Footnote 102 But in-principle enforceability here cannot just mean that the court has “standing” to enforce an obligation, since a court lacks such standing, legally and morally, if the claim is improperly filed or insufficiently evidenced. So in-principle enforceability must be understood as hypothetical enforceability:
(2) N is a legal norm in C if and only if it is a moral norm that is judicially enforceable in C or C’.
(2) counts a right as legal even if it is enforceable only counterfactually, i.e., only in a possible world (where C’ obtains). Notice that it is no longer straightforwardly an account of enforceability “on demand.” It might nonetheless give acceptable results for the procedural and evidentiary doctrines: it can suppose that in C’, A's claim is properly filed and sufficiently evidenced.
Of course, in some possible world, practically any conceivable moral norm is enforceable in court. LE cannot count any such norm as legal: simply fictionalizing will not tell us about legal rights in the actual world. So there must be a closeness restriction on C’. The problem is that if the closeness restriction is plausible, LE will give implausible results; if, however, the closeness restriction is permissive enough to give the right results, LE loses intrinsic plausibility and explanatory power, and seems ad hoc.
Consider doctrines of nonjusticiability (sometimes called political question doctrines). Although courts sometimes appeal to these doctrines as an imprecise label for the nonexistence of a legal duty, they are also used more accurately to indicate that a court may not adjudicate genuine legal duties, such as certain duties governments owe in armed conflicts, or when acting in coordination with foreign states.Footnote 103 These claims are not enforceable in court in the actual world,Footnote 104 or in any nearby possible world, which will also contain the nonjusticiability doctrine. For these doctrines are not a minor or incidental feature of the actual circumstances but, where they exist, a fundamental part of the legal structure, delineating the institutional role of courts and affecting the content of substantive legal rights and duties (if there were no nonjusticiability doctrine, the content of government duties might well be more permissive).
The problem is even clearer in international law, an extensive body of legal rights and duties, many of which—without any court of compulsory, general jurisdiction—cannot be enforced. LE cannot explain how these are legal rights unless it allows that C’ need not be reasonably close: these rights would be enforceable only in a world with a court of compulsory, general jurisdiction.Footnote 105 But that world is not only very distant from ours politically; it is one in which it would make no sense to enforce our legal rights and duties. Think of states’ legal rights to take “countermeasures,” certain otherwise unlawful acts that are lawful if taken in response to another state's prior breach of an international obligation. This legal right reflects—both causally and normatively—the need for certain forms of “self-help” in an international system without centralized enforcement. LE can explain its existence only if it assumes away the very conditions that make it intelligible. In each case, (2) cannot account for legal rights, unless it appeals to an implausible hypotheticalization. And if such distant possibilities are allowed to count as the relevant counterfactual in (2), there seems no principled limit to what moral rights will be counted as legal.
Is there a better, more explanatory notion of in-principle enforceability available to LE? Here is one possibility:Footnote 106
(3) A norm, N, is a legal norm in circumstances C if and only if it is a moral norm that is either (a) judicially enforceable in C or (b) not enforceable in C because a different legal norm, O, requires or permits the court not to enforce it.
(3) ties in-principle enforceability to the presence of a further legal norm, which explains the first norm's actual unenforceability. It deals with instances where, say, a legal norm is nonjusticiable because the case raises a political question. (b) will here be satisfied: the norm (N) is unenforceable because the nonjusticiability rule (O) permits or requires the court not to enforce it.
However, (3) fails to capture at least some legal duties whose unenforceability is not due to the operation of norms of the O type. For example, highest courts in many jurisdictions have legal duties to hear some appeals, decide cases argued before them, and so on. These legal duties are generally unenforceable, but not by virtue of the operation of some further norm permitting or requiring their nonenforcement. The same is true of norms of international law whose unenforceability is simply due to the absence of any relevant court.
More fundamentally, (3) is over-inclusive because (3)(b) is too permissive. Think of moral rights (N) generated by precontractual commitments or ministerial statements, where a judicial decision, statute, or constitutional provision (O) specifies that precontractual negotiations or legislative history do not give rise to contractual or statutory rights. (3) wrongly counts these moral rights as legal because (3)(b) is satisfied. The same is true even more clearly of statutes (O) which repeal others (N), where there is still a moral norm with the same content as the repealed law (e.g., a legal duty of care is repealed but a moral duty remains). Here the repealing laws cause these norms, which otherwise would have been enforceable, to be unenforceable. Yet despite satisfying (3)(b), they are of course not legal.
Unlike true justiciability norms, the O norms in these examples operate by preventing the N norms from arising as legal rights in the first place, or by extinguishing their legal status. A proponent of LE might therefore want to limit (3)(b) so that it only applies where the underlying N norms, though unenforceable, are legal norms nonetheless. But LE lacks the resources to make such a limit coherent. Consider:
(4) A norm, N, is a legal norm in circumstances C if and only if it is a moral norm that is either (a) judicially enforceable in C or (b) not enforceable in C because a different legal norm, O, requires or permits the court not to enforce N, although N is a legal norm.
(4)(b) is circular. It makes legality an irreducible part of the definition and explanation of one species of legality. (4)(b) is meant to explain a way for unenforceable moral norms to nevertheless count as legal, but the proposed condition these norms must meet assumes they are somehow independently legal. It presupposes its own incompleteness.
(3)'s over-inclusiveness cannot, then, be cured by moving to (4). And other problems of over-inclusiveness apply to LE generally, given its implication that every moral norm properly enforced by a court is a legal norm. For example, some cases require the judicial enforcement of foreign law, and hence potentially moral obligations that are not legal obligations within the jurisdiction. So LE is false if indexed to a single legal system, since it will say that legal obligations within that system include obligations under foreign law. Yet LE must in some way be indexed to a particular legal system, since legal obligations come in systems and differ between them.
Consider, too, that according to a familiar and attractive picture of judicial decision-making, courts have standing to enforce some moral obligations that are not legal obligations prior to being judicially recognized—to incrementally extend the reach of legal norms. Presumably, some of these extensions recognize in law what are already moral obligations (e.g., not to negligently injure certain persons). In these cases, the “standing” version of LE identifies a legal obligation prematurely—before the relevant judicial decision—since there is, ex ante, already a moral obligation that the court has standing to enforce. Other variants of LE have the same problem where a plaintiff ought to, but does not yet, have their moral right legally protected by the court.Footnote 107 A proponent of the one-system view might reject the familiar picture of judicial decision-making and insist that these apparently “premature” legal rights are correctly counted as valid. But this would not be costless, either in terms of extensional plausibility or theoretically. It supposes that courts cannot properly extend legal rights; they simply recognize what were already legal rights to begin with. This response would therefore saddle the one-system view with a highly controversial, broadly Dworkinian theory of adjudication that is not required by the one-system view as such.
If neither MI nor LE is sound taken by itself, might they be combined into a stronger, hybrid set of demarcation criteria? A hybrid might be either conjunctive or disjunctive. On a conjunctive hybrid, for a moral norm to be legal it must both have been generated in the legally proper way and be enforceable in court. But such a conjunction will simply incorporate the under-inclusiveness that was the main problem with LE. A disjunctive hybrid view would hold that, for a moral norm to be legal, it must satisfy either MI or LE. Yet this suffers from the opposite problem. It incorporates the over-generation of legal norms of LE and, in particular, MI.
V. OBJECTIONS
The Identity Thesis, I have suggested, is not tenable. It implies the Correspondence Thesis, which is false, and it delivers under- and over-inclusive results when spelled out with possible demarcation criteria. Let me now consider two possible responses to these arguments.
A. Heuristics
Hershovitz has argued that despite our willingness to think and speak of legal obligations that do not correspond to moral obligations, this does not reflect the existence of such legal obligations, but rather our use of heuristics. Sometimes, Hershovitz says, we should be “morally obtuse about our moral obligations,”Footnote 108 construing them as more absolute than they really are. Parents should treat their responsibilities to love and support their children as unconditional even though, if they were to “think about things in a clearheaded fashion,” they would see that “of course” they are not.Footnote 109 Likewise, claims Hershovitz, in the law: we should use certain heuristics—e.g., that duly enacted statutes should always be enforced—for determining when moral-legal duties exist, even if the moral reality is more complicated than what the exceptionless heuristic contemplates. For we would not want legal officials to “think of themselves as moral arbiters of the acts of Congress”; instead they should have “humility and deference,” which the simple heuristic helps them achieve.Footnote 110
Should we really employ such uncompromising heuristics in thinking about law? For these heuristics to explain ordinary—not just official—thought, we would all have to be using them—and for ordinary citizens, “humility and deference” toward legal institutions seem naïve or dangerous attitudes. No heuristic should assign to the actions of legal institutions a moral relevance that, as we have seen, they systematically lack.
Even if we should use these heuristics, it does not follow that we do use them, which is what the argument must show to explain our actual thought and talk. And that descriptive claim does not seem generally true: very many people think the law is an ass, or are just ambivalent about its moral character, and so do not assume that statutes or judicial decisions always reflect or give moral obligations.
Most importantly, even if we did employ such heuristics, that would mean we thought and spoke as if there were legal obligations in cases where in fact there were none (for lack of corresponding moral obligations). By parity of reasoning with the parent case, we should then be able to see, when we reflect clearheadedly, that we lack the legal obligations in question. But the lesson of Sections II and III was to the contrary: in many familiar cases where one lacks a moral obligation, we recognize clearly the existence of a legal obligation.
B. Begging the Question?
It might be claimed that putative counterexamples to the one-system view are question-begging.Footnote 111 But in what way? The argument does not assume the truth of legal positivism, and, as we saw in Section I, denying the Identity and Correspondence Theses is consistent with various jurisprudential positions, as is accepting the examples canvassed as possible legal norms. As that earlier discussion also showed, there is nothing question-begging in asking whether legal and moral norms correspond, for correspondence, in the relevant sense, is just for them to have the same content—and that is a necessary condition for the one-system view to hold.
Does suggesting that certain examples are counterexamples to the one-system view illegitimately presuppose its falsity? No. There is nothing unusual or illegitimate in the argumentative strategy here—a form of reductio, or refutation by counterexample, which is employed in many philosophical contexts. The standard exemplar is Gettier's celebrated attack on the justified, true belief theory of knowledge. His argument highlighted cases where someone might have a justified, true belief but not knowledge.Footnote 112 The argument pursued here is structurally parallel. The analogy is inexact, obviously—not only in the dimension of philosophical ingenuity, but because the force of Gettier's examples was largely uncontroversial, whereas the examples in Sections II–IV might be disputed by proponents of the one-system view. Nevertheless, the success of a reductio does not rely on adherents of the view it targets accepting its success. That could not sensibly be insisted on as a standard for philosophical argument.Footnote 113
Reductio arguments are legitimate and frequently important even where their conclusions are not uncontroversial. Consider: according to a simple kind of internalism about reasons, one has a reason to φ only if φing would further one's desires. It would be neither question-begging nor immaterial to reply, by way of reductio, that this falsely implies that a person thoroughly committed to murdering has no reason not to murder. Nor is it question-begging when critics of consequentialism argue that it absurdly implies that worlds in which some number suffer massive pain are preferable to those in which some much larger number experience trivial discomfort. Of course, there are many reasons why these particular criticisms might not stick, and many ways their targets might reply. But they are not illegitimate in form, and do not become question-begging because some reasons internalists or consequentialists reject them. The reluctance to countenance an implausible result does not make a theory dialectically immune to counterexamples.Footnote 114 Nor, by the same token, does preemptively accepting that one's theory may have “counterintuitive consequences.”Footnote 115
In all reductio arguments, some putatively implausible result counts as implausible only if the target theory is unsound. But the target theory's unsoundness is an implication, not a presupposition, of the argument. Accordingly, my strategy does not first assume the falsity of the one-system view and thereby derive particular conclusions about legal and moral norms. Instead, it moves from (inter alia) particular claims about legal and moral norms to show the falsity of the view. This is simply modus tollens; schematically: (1) the one-system view implies p (certain arrangements of legal and moral norms); (2) p is false; (3) therefore, the one-system view is false. If we construe the argument in more epistemically modest terms—so that it concludes only in the probable falsity of the one-system view, or in a reason to reject itFootnote 116—the order of inference is unchanged.
It is no objection that premises (1) and (2) rely partly on judgments about examples (they also rely on more abstract considerations). Proponents of the one-system view themselves rely on such judgments: Dworkin's famous argument from theoretical disagreement appeals to judgments about legal practice;Footnote 117 Greenberg derives support from putatively intuitive or natural judgments about the legally proper way of creating obligations,Footnote 118 among other things;Footnote 119 Schaus thinks a theory “won't do” if it “recommends conclusions . . . that we confidently reject.”Footnote 120 My point is not a tu quoque. It is that, as such views rightly suggest, one cannot assess a theory's adequacy to its explanandum without making judgments about the explanandum's properties, in general and in particular cases. That is as true of legal obligations as it is of other objects of philosophical inquiry. These judgments can of course be disputed. It might be argued that the moral analysis in Sections II–IV is mistaken, so that the one-system view does not have implausible upshots about legal content. This response would have to be pursued in detail, case by case. A proponent of the one-system view might instead bite the bullet and agree that their view has the implications about legal content I suggested. By itself, that would simply be to accept rather than answer these extensional problems. In any event, the argument here cannot be dismissed wholesale as question-begging, or on the ground that its conclusions would not matter even if true.
VI. CONCLUSION
If the argument here succeeds, what follows? Notice, first, what the argument suggests about the normativity of law. While that topic has not been my direct focus, the argument here turns out to have a significant bearing on it. It suggests that the prospects for subsuming legal normativity into morality—or some other “robust” (genuinely reason-giving) form of normativity—are not good. For it will not be possible to substantiate the necessary correspondence between legal norms and those of morality or another similar domain. We identify, understand, and reason with legal norms without regard to such correspondence—and no less so when it does not hold. If that is right, and legal normativity is not subsumable into morality, or otherwise robust normativity, then we have at least one example of a domain of normativity that functions, in some sense, nonrobustly. Obviously these remarks do not amount to a full account of legal normativity, and they leave much to be spelled out. But they do identify a significant aspect of legal normativity that any successful account of it should explain.
That is the indirect implication of the critique in this paper. Its direct implications concern anti-positivism's prospects and possible forms. As I noted, the one-system view represents the dominant approach among contemporary anti-positivists. If the critique here is sound, then anti-positivism has taken a serious wrong turn. In the form in which it is most widely endorsed today, anti-positivism appears unviable.
Yet the critique here also points toward a possibly more fruitful direction. Proponents of the one-system view might be persuaded that it is not the best way of elaborating anti-positivism. Many of the theoretical arguments for anti-positivism—though they also face serious challenges—do not entail the one-system view in particular. Greenberg's arguments that legal content must be determined “rationally”Footnote 121 do not entail it; nor do Dworkin's arguments from the nature of interpretation or theoretical disagreement in law.Footnote 122 Although such arguments posit an essential role for moral considerations in grounding legal norms, they are quite consistent with the picture of legal normativity just adumbrated. In short, one might hold that one can get from social facts to legal facts only via some intervening moral explanation, yet deny that the legal facts thereby explained are themselves just more moral facts. Anti-positivists could abandon the Identity Thesis.
Alternatively, the one-system view's proponents might cast it as a revisionary project—one that seeks not to explain legal obligations, rights, and powers as we now understand them, but to recommend a project of conceptual reform, a new way of carving up the social and political world.Footnote 123 The risk of quietism, touched on in discussing Hershovitz's heuristics argument, would still apply to this proposal, but the critique here does not show conclusively that this route is not viable. It does suggest, however, that such a revisionary project would involve a deep departure from our present understanding of law.
